Research On Automatic Emergency Braking Model Based On Multilayer Perceptron Under Severe Weather Environment

Autonomous Emergency Braking(AEB)system is the basic safety module for intelligent vehicles traveling in complex road traffic environment.It is a research hotspot in the field of intelligent automobile at home and abroad.This paper addresses the key technologies of autonomous emergency braking system,analyses the working state of vehicle active safety systems in severe weather environments,constructs a multi-layer perceptron model,and carries out research on severe weather environment accident severity prediction methods and algorithm optimization of adaptive automatic emergency braking systems,mainly including(1)Processing and analysis of severe weather environment and traffic accident data setsBased on the severe weather environment and traffic accident datasets,the contents of the datasets were analysed in detail to confirm the reliability and practicality of the datasets,and the training data and validation data for the multi-layer perceptron model were provided through data cleaning of the datasets.(2)Building a multi-layer perceptron to predict the severity of traffic accidentsBased on the training and validation datasets,the multilayer perceptron model is constructed by considering the non-linear correlation between the data,adjusting several hyperparameters of the multilayer perceptron model to improve the accuracy of the model and reduce the misjudgement rate of the model,so as to provide prediction capability of traffic accident severity under severe weather conditions.(3)Establishing and optimising adaptive emergency braking system algorithmsBased on the accident severity prediction capability provided by the multilayer perceptron,the automatic emergency braking algorithm is constructed and optimised to build an adaptive emergency braking algorithm that can adapt to a variety of adverse weather conditions(4)Experimental verification based on virtual simulation and driver in the loopThe adaptive emergency braking algorithm is validated based on the Prescan simulation platform and the driver-in-the-loop platform,and the performance of the adaptive emergency braking system algorithm is compared with that of the emergency braking system algorithm by setting up simulation scenarios in various environments to verify the feasibility and reliability of this research.